SLAPBench, a new 7,832-pair benchmark, shows prompt format controls MLLM collapse on SLAP fingerprint verification and that a perfect AUC from Qwen3-VL-8B is a near-duplicate artifact, not identity capability.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2026 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
SLAPBench: Benchmarking Multimodal Large Language Models for Four-Finger SLAP Fingerprint Verification
SLAPBench, a new 7,832-pair benchmark, shows prompt format controls MLLM collapse on SLAP fingerprint verification and that a perfect AUC from Qwen3-VL-8B is a near-duplicate artifact, not identity capability.